Pneumonia Detection in Chest X-Ray Images : Handling Class Imbalance

Pneumonia affects individuals all around the world, although the highest fatality rates occur in Sub-Saharan Africa and South Asia. Despite immunizations and medications, pneumonia incidence and fatality rates have increased, highlighting the need for improved prevention and treatment. Researchers have explored frameworks for reliable pneumonia detection, diagnosis, and analysis, but AI progress is hampered by a lack of sufficient and balanced data. In this paper, we provide a novel framework for addressing class imbalance that employs a mix of Deep Convolutional Generative Adversarial Network (DCGAN) and Wasserstein GAN with gradient penalty (WGAN-GP) for minority class augmentation and Random Under-Sampling (RUS) for majority class. Validation on the massive ChestX-Ray8 dataset using transfer learning on ResNet-50, Xception, and VGG-16 produces results that outperform existing benchmarks.